The Backup Conundrum: More Data in Less Time Page 2


Want the latest storage insights?

Download the authoritative guide: Enterprise Data Storage 2018: Optimizing Your Storage Infrastructure

Option Two: Consolidate Data onto a SAN

Another way to realize the goal of backing up more data in less time is to consolidate data onto a SAN. Peter Hunter, EqualLogic’s product marketing manager, maintains that moving data to a SAN not only reduces storage management overhead, but also allows for a plethora of backup solutions that result in reduced backup windows and increased service levels.

The key, says Hunter, is that the disk arrays, tape drives, application servers, and backup servers become connected over a storage network. Hunter says he personally recommends iSCSI SANs because of their low cost and familiar IP infrastructure, which makes the task of connecting all of the servers to a SAN much easier.

Disk-to-Disk or Disk-to-Disk-to-Tape?

A third way to achieve the goal, according to Sante, is by using disk-to-disk-to-tape backup. He says IT managers can back up to disk at 100 Megabytes per second (MBps), versus backing up to tape drives at 1 to 12 (MBps), and then archive that disk-based backup to tape.

“Disk storage systems are at least ten times faster than tape drives and don’t require streaming data to maintain that performance,” says Sante. “To decrease the backup window, users should back up files to disk systems.”

His rationale is based on the fact that servers burst data, making it difficult for tape drives to run smoothly due to their need for continuous data streams. By first backing up to disk you can eliminate the bottleneck of tape drives and back up files at disk speeds instead of tape speeds.

“With the price of disk systems dropping and the introduction of IP-based SANs, it is now very affordable to use a shared disk storage system as the primary backup media for multiple servers and then to use tape as the archive media. Most backup software packages support disk-to-disk-to-tape options,” says Sante.

Hunter recommends taking it a step further, by cutting off tape backup entirely whenever possible. He feels the performance of tape drives and libraries is a major bottleneck for backup and suggests implementing straight disk-to-disk backups. “Disk is cheaper and more reliable than tape,” says Hunter.

“By backing up to disk, backup times accelerate to disk speeds,“ he says. However, he admits that most organizations are not ready to abandon their tape so quickly, and tape has certain durability and portability advantages over disk, to say nothing of cost. The good news is that, according to Hunter, disk backups are not necessarily a replacement for tape backups. “By staging backups to disk and then to tape, administrators streamline the backup process and create an additional backup resource,” he says.

Hunter explains that the actual transfer to tape is accelerated because the application server is not involved in that stage, and avoids the “shoe-shining” problem associated with tape backups. “For most failures, administrators can recover from this disk repository, increasing recovery times,” he says.

Page 3: Implement Proxy-based Backups

Submit a Comment


People are discussing this article with 0 comment(s)